Hi, The problem is well related to ARP traffic. During the problem, please run few times the 2 commands below, to get a sampled base: sh clock sh ip traffic | b ARP   By the way, your platform is older than Gandalf.   Remi Astruc ... View more

Hello, Can you please issue the following commands and give the outputs: show ver show ip traffic show int | i Input queue|line protocol is up   Remi Astruc ... View more

Hi, Good. About the DCs, if the requirement is to perform DHCP to the Infoblox, that can be managed from within the BD by assigning a DHCP Relay policy towards the Infoblox behind L3Out, and you're done. Mark the answer as solved if that's fine for you. Remi Astruc ... View more

Hello, Yes, but then you cannot connect the Domain Controllers on the Leaves anymore. You will need to provide the L2 Vlan with additionnal switches connecting L3out ports, Infoblox and DCs. But what is really your requirement of having DCs on the same Vlan as the Infoblox? Maybe that can be solved differently.   Remi Astruc ... View more

https://community.cisco.com/t5/application-centric/aci-l3-out-connectivity-doubt/m-p/3818236/highlight/true#M6197   ... View more

Hello, No. By nature, L3Out is precisely anything but a Bridge Domain. You will also never establish OSPF peering with the Fabric from within a BD. If your Infolblox is a dynamic routing instance in your design, you will put it on a L3Out. You can still route traffic between the L3Out routers and the BD Endpoints if needed, with an appropriate contract.   Remi Astruc ... View more

Hello, snmp-server trap-source gigabitEthernet 0/0 snmp-server host x.x.x.x vrf MGMT   Remi Astruc ... View more

Hello, No. In Stretched Fabric design, the uplinks between Leaf and Spine can only be 40G or 100G (and 400G is coming). DWDM can be used to carry natively the 40G uplinks (or 100G/400G), or can be used to carry an MPLS Pseudowire of 10G but in that case the 10G ends on the MPLS router, not on the Leaf/Spine.   Remi Astruc ... View more

Hello, You could create a VRF or use an existing one (in case you need someday to integrate in routing), that's no big deal. Regarding the creation of 1 BD for several isolated Vlans made with EPGs, I would not do that way. Even if no contract, the EndPoints could still be visible to some traffic like ARP or L2 Uknown Unicast set to Flood (as recommended in L2 BD).   Remi ... View more

Hi Fabio,   I would really recommend not to do that. There are numerous drawbacks and limitations.   You will need to disable L3 Configuration and Unicast Routing in the Bridge Domains. Then you will manage/view all your Endpoints with MAC addresses only. The Fabric will not enhance Endpoint detection by ARPing destinations. Let's assume you have 2 Subnets/BDs A and B. You'll need to create an EPG for SVI A. But because of inter-SVI routing, that EPG will also represent all the Endpoints located in Subnet B, which are in reality in other EPGs attached to BD B... Setting a contract between EPG Subnet A and EPG Subnet B will just not work. You'll need EPG A to SVI A, then SVI B to EPG B. Routing outside the Fabric should really be avoided except using L3Outs. Your SVI devices will be the bottleneck of your network and will break the scale-out concept of distributed routing. Forget about VRF Route-Leaking, PBR, ... And so on... I'm not even talking about the financial impact of using an ACI Fabric with 80% of features unused.   Hope you'll convince your managers for the sake of your network team!   Remi Astruc ... View more

Hello,   That is possible! You just need to unveil the hidden menu by actually selecting the Symmetric Hashing Control:   Remi   ... View more